Article

Cite This: Biochemistry 2018, 57, 5513−5523 pubs.acs.org/biochemistry

Analysis of the Cysteine Residue of the Sacrificial Sulfur Insertase LarE from Lactobacillus plantarum † ⊥ † ‡ † § † ∥ Matthias Fellner, , Joel A. Rankin, Benoît Desguin, Jian Hu, , and Robert P. Hausinger*, , † Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, Michigan 48824, United States ‡ Institute of Life Sciences, Universitécatholique de Louvain, B-1348 Louvain-La-Neuve, Belgium § Department of Chemistry, Michigan State University, East Lansing, Michigan 48824, United States ∥ Department of Microbiology and Molecular Genetics, Michigan State University, East Lansing, Michigan 48824, United States

*S Supporting Information

ABSTRACT: LarE from Lactobacillus plantarum is an ATP- dependent sulfur that sacrifices its Cys176 sulfur atom to form a dehydroalanine (Dha) side chain during biosynthesis of the covalently linked nickel-pincer nucleotide (NPN) (pyridinium 3-thioamide-5-thiocarboxylic acid mononucleotide) of lactate racemase. (CoA) stabilizes LarE and forms a CoA-Cys176 mixed disulfide with the protein. This study presents the crystal structure of the LarE/CoA complex, revealing protein interactions with CoA that mimic those for binding ATP. CoA weakly inhibits LarE activity, and the persulfide of CoA is capable of partially regenerating functional LarE from the Dha176 form of the protein. The physiological relevance of this cycling reaction is unclear. A new form of LarE was discovered, an NPN-LarE covalent adduct, explaining prior results in which activation of the lactate racemase apoprotein required only the isolated LarE. The crystal structure of the inactive C176A variant revealed a fold essentially identical to that of wild-type LarE. Additional active site variants of LarE were created and their activities characterized, with all LarE variants analyzed in terms of the structure. Finally, the L. plantarum LarE structure was compared to a homology model of Thermoanaerobacterium thermosaccharolyticum LarE, predicted to contain three cysteine residues at the active site, and to other proteins with a similar fold and multiple active site cysteine residues. These findings suggest that some LarE orthologs may not be sacrificial but instead may catalyze sulfur transfer by using a persulfide mechanism or from a labile site on a [4Fe-4S] cluster at this position.

actate racemase (Lar) interconverts the D and L isomers of from Thermoanaerobacterium thermosaccharolyticum (Uni- L lactic acid, a central metabolite of many microorganisms.1 ProtKB entry D9TQ02) does not require adduct formation.8 The allows for the production of D-lactate for cell wall This study focuses on LarE, a member of the PP-loop biosynthesis in bacteria that possess only L-lactate dehydrogen- pyrophosphatase family containing a PP-loop SGGxDS motif Downloaded via MICHIGAN STATE UNIV on September 25, 2018 at 12:31:05 (UTC). ase and permits the metabolism of both isomers from racemic in its N-terminal region. Transformation of pyridinium 3,5- mixtures of lactate in microorganisms containing a single form biscarboxylic acid mononucleotide (P2CMN) to pyridinium See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. 2−4 of lactate dehydrogenase. The Lar enzyme from Lactoba- 3,5-bisthiocarboxylic acid mononucleotide (P2TMN) by LarE cillus plantarum includes the protein LarA (UnitProtKB entry requires two cycles of carboxylate activation (involving ATP- F9USS9) and a covalently tethered (via Lys184) cofactor, dependent adenylylation) and sulfur insertion, with the sulfur pyridinium 3-thioamide-5-thiocarboxylic acid mononucleotide originating from a cysteine residue (Cys176) of the protein, with nickel bound to C4 of the pyridinium ring, the two sulfur thus making LarE a sacrificial sulfur transferase that acquires a atoms of the pincer complex, and the His200 side chain 8,9 dehydroalanine (Dha) residue. (Figure 1).5,6 This organometallic complex will subsequently Here, we examine several unusual and interesting features of be termed the nickel-pincer nucleotide (NPN) cofactor.7 The LarE with focus on its Cys176 residue. First, we investigate the NPN cofactor is synthesized from nicotinic acid adenine dinucleotide (NaAD) by the sequential actions of LarB unresolved nonsubstrate interaction of LarE with coenzyme A (UniProtKB entry F9UST0),8 which carboxylates C5 of the (CoA), a cellular component that stabilizes LarE and forms a disulfide with Cys176 but is not believed to be required for pyridinium ring and hydrolyzes the phosphoanhydride bond, 8,9 LarE (UniProtKB entry F9UST4), which converts the two LarE activity, by presenting the crystal structure of LarE with carboxylates to thiocarboxylates,9 and LarC (UniProtKB entry F9UST1), which installs the nickel.8,10 The process by which Received: May 29, 2018 the NPN cofactor becomes covalently attached to LarA from Revised: August 28, 2018 L. plantarum has not been described, and activation of LarA Published: August 29, 2018

© 2018 American Chemical Society 5513 DOI: 10.1021/acs.biochem.8b00601 Biochemistry 2018, 57, 5513−5523 Biochemistry Article

pentaerythritol ethoxylate (15/4 EO/OH), 50 mM Bis-Tris (pH 6.5), and 100 mM ammonium sulfate. The CoA-bound sample contained 28 mg/mL WT LarE mixed with 0.9 mM CoA and was incubated for ∼10 min on ice before setting up the drop. The C176A LarE sample contained 8 mg/mL variant protein. Using the same setup that was used for C176A, we also set up unsuccessful crystal drops for W97A (∼1 mg/mL), S180A (∼10, 12, and 15 mg/mL), R212A (∼6 mg/mL), and D231R (∼3, 11, and 24 mg/mL) variants of LarE using the same procedures.9 Diffraction Data Collection, Structure Determination, and Analysis. Data sets were collected at the Advanced Photon Source LS-CAT beamlines (21-ID-F and 21-ID-G). Data sets were processed with xdsapp11 and iMos,12 with merging and scaling done using aimless.13 Phaser molecular replacement14 was utilized using the WT apoprotein model fi Figure 1. NPN cofactor in lactate racemase. Lys184 of LarA binds 5UDQ. Model building and re nement were conducted in Coot15 and Phenix.14 Data set statistics are listed in Table 1. pyridinium 3-thioamide-5-thiocarboxylic acid mononucleotide 16 [shown as sticks with cyan carbon atoms; Protein Data Bank UCSF Chimera was used to create structure figures. (PDB) entry 5HUQ]. Two sulfur atoms and one carbon atom of the Purification of Dha-Containing LarE. LarEDha was cofactor along with His200 coordinate nickel (green sphere). purified from Lactococcus lactis NZ3900 containing pGIR172 that carries the larA−larE genes from L. plantarum with larE bound CoA. Second, we provide evidence that CoA weakly translationally fused to DNA encoding the Strep-tag II5 as inhibits LarE activity. Third, we demonstrate the ability of previously described.8 The cells were grown overnight without CoA-persulfide to regenerate Cys-containing LarE from the being shaken in 15 mL of M17 medium (Oxoid or Difco) Dha Dha176-containing protein (LarE ). Fourth, we provide containing 0.5% (w/v) D-glucose and 10 mg/L chloramphe- evidence of the presence of a novel NPN adduct of LarE, nicol at 30 °C. This inoculum was added to 1.5 L of the same providing an explanation for earlier confounding observations medium, but containing 5 mg/L chloramphenicol, and related to the ability of purified LarE to activate the LarA incubated at 28 °C while being shaken (40 rpm) until the ∼ apoprotein. Fifth, we investigate whether the loss of the OD600 reached 0.3. The culture was supplemented with 1 fi μ Cys176 side chain signi cantly impacts the protein fold by mM NiCl2, induced with 10 g/L nisin A (Sigma), and grown characterizing the structure of the C176A variant. Sixth, we for 4 h before the cells were harvested by centrifugation at characterize the functions of several other key residues of LarE 5000g for 10 min. The cell pellet was washed using 100 mM and summarize all LarE mutagenesis studies to better define Tris-HCl buffer (pH 7.5) containing 300 mM NaCl and stored the sacrificial sulfur insertion reaction.9 Lastly, we speculate until it was needed at −80 °C. The thawed cells were about whether LarE homologues containing additional active suspended in 40 mL of 100 mM Tris-HCl buffer (pH 7.5) site cysteine residues may operate by a catalytic sulfur transfer containing 300 mM NaCl, 1 mM phenylmethanesulfonyl process. fluoride (PMSF, added as a 100 mM stock in ethanol), and 1 unit/mL Benzonase (Millipore) on ice and then lysed by two ■ MATERIALS AND METHODS passes through a chilled French pressure cell at 16000 psi. The Genes, Plasmids, and Cloning. Site-directed mutagenesis debris and intact cells were removed by centrifugation at of the gene encoding Strep II-tagged wild-type (WT) LarE 27000g and 4 °C for 40 min. Purification of LarEDha was from L. plantarum was performed using a QuikChange performed by using a 1 mL bed volume of Strep-tactin-XT mutagenesis kit (Agilent) for creating the K101A, E128A, (IBA) resin equilibrated in 20 mM Tris-HCl buffer (pH 7.5) and E223A variants. All of the constructs were verified by DNA containing 300 mM NaCl at 4 °C. The lysates were loaded at a sequencing. Details, including instructions for creating the rate of 1 mL/min; the resin was washed by gravity flow with C176A variant, are summarized in our previous work.9 five 1 mL additions of 20 mM Tris-HCl buffer (pH 7.5) LarE Overexpression, Purification, and Character- containing 300 mM NaCl, and the protein was eluted with six ization. WT, C176A, and the new variant forms of LarE 0.5 mL additions of 20 mM Tris-HCl buffer (pH 7.5) were overexpressed and purified using our previously described containing 300 mM NaCl and 50 mM biotin (Santa Cruz pGIR076 Escherichia coli ArcticExpress-Strep-tactin (IBA) Biotechnology) that had been neutralized with 50 mM NaOH. system.9 The final buffer for all samples was 100 mM Tris- Additional characterization of LarEDha included analysis of HCl (pH 7.5) containing 300 mM NaCl. The activities of WT its reactivity with 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB) and variant forms of LarE were assessed by monitoring samples and tris(2-carboxyethyl)phosphine (TCEP) in comparison to for their abilities to transform P2CMN to P2TMN that was that of LarE. To quantify accessible thiol groups, the protein further metabolized by LarC and incorporated into LarA samples were treated with 180 μM DTNB in 100 mM Tris- apoprotein from T. thermosaccharolyticum; this activated LarA HCl (pH 7.5), allowed to react for 30 min, and monitored at enzyme was then assayed for lactate racemization as described 412 nm.17 For testing the reactivity of LarE-associated Dha elsewhere in detail.9 with TCEP,18 the samples were treated with 5 mM TCEP in Crystallization. For both the CoA-bound WT protein and buffer and analyzed by mass spectrometry (see below). the C176A LarE variant, crystals were obtained after mixing 5 LarE Regeneration Reactions. Lysates for use in LarE μL of protein samples with 5 μL of reservoir solutions. The regeneration experiments were derived from cultures of L. hanging drop reservoir contained 100 μL of 30.0% (v/v) plantarum NCIMB8826. Using a 2% inoculum, the cells were

5514 DOI: 10.1021/acs.biochem.8b00601 Biochemistry 2018, 57, 5513−5523 Biochemistry Article

Table 1. Crystallography statistics for the CoA-Bound and with two 60 s pulses separated by 5 min on ice. The debris was C176A Variant LarE Structures removed by centrifugation at 20000g and 4 °C for 10 min. (IscS) was purified from E. coli BL21 CoA-bound C176A (DE3) containing pBH402, a plasmid with E. coli iscS fused to Data Collectiona a sequence encoding a His6 tag and under the control of the beamline LS-CAT 21-ID-F LS-CAT 21-ID-G T7 promoter.20 Overnight cultures of these cells were grown wavelength (Å) 0.979 0.979 while being shaken (300 rpm) at 37 °C in LB containing 50 space group P4122 P4122 μg/mL kanamycin (Gold Biotechnology), then diluted 200- unit cell dimensions fold into 0.5 L of Lennox LB supplemented with 1 mM a, b, c (Å) 107, 107, 320 109, 109, 329 pyridoxine (Sigma) and 50 μg/mL kanamycin, and further α β γ , , (deg) 90, 90, 90 90, 90, 90 grown while being shaken at the same temperature. When an resolution (Å) 47.95−2.09 48.67−2.35 ∼ (2.12−2.09) (2.39−2.35) OD600 of 0.5 was reached, the cultures were amended with β D no. of unique reflections 111492 (5293) 82741 (4345) 0.4 mM isopropyl - -1-thiogalactopyranoside and grown for an additional 3 h. The cells were collected by centrifugation redundancy 6.1 (6.0) 11.3 (11.2) ff completeness (%) 99.7 (97.1) 99.5 (97.3) and resuspended in 40 mL of bu er A [50 mM Na2HPO4 (pH I/σI 14.7 (1.8) 18.9 (2.1) 8.0) containing 300 mM NaCl and 20 mM imidazole] b supplemented with 2.5 units/mL Benzonase and 1 mM Rmerge 0.084 (0.899) 0.102 (1.079) c PMSF. The suspension was chilled on ice and lysed by two Rpim 0.055 (0.599) 0.046 (0.482) d passes through a chilled French press cell at 16000 psi. The cell CC1/2 0.998 (0.509) 0.999 (0.702) ° Refinement debris was pelleted by centrifugation at 27000g and 4 C for 40 no. of protein atoms 11949 11420 min. The supernatant was applied to a column (10 mL) of Ni- nitrilotriacetic acid-Sepharose that had been equilibrated in no. of CoA atoms 288 0 ff no. of phosphate atoms 30 30 bu er A. After the column had been washed twice with 50 mL of buffer A, IscS was eluted at a rate of 2 mL/min with a 100 no. of sulfate atoms 0 25 ff ff no. of waters 606 493 mL linear gradient from bu er A to bu er A containing 500 R /R e 0.195/0.234 0.187/0.244 mM imidazole. The 2.5 mL fractions were collected into tubes work free μ B factor (Å2) 43.8 56.3 containing 2.5 L of 10 mM EDTA (pH 8.0) to prevent protein 43.9 57.0 precipitation. The IscS-containing fractions were combined ° CoA 59.5 − and dialyzed overnight at 4 C against 2 L of 50 mM potassium ff phosphate 38.5 47.7 phosphate bu er (pH 7.5) containing 5 mM MgCl2, 100 mM sulfate − 61.5 KCl, and 0.1 mM EDTA. The protein concentration was fi ε −1 −1 20 water 42.5 48.4 quanti ed using the published 280 of 25400 M cm , and fi ° root-mean-square deviation in bond 0.007 0.007 puri ed IscS was stored at 4 C. lengths (Å) Organic persulfides were synthesized by incubating 5−20 root-mean-square deviation in bond 0.975 0.922 mM disulfides with a 5-fold molar excess of NaHS in 300 mM angles (deg) Tris-HCl (pH 7.5, for making CoA and glutathione Ramachandran plot, favored (%) 97.34 97.43 persulfides) or 100 mM NaOH (for the Cys persulfide) in Ramachandran plot, outliers (%) 0 0 an argon atmosphere.21 The reaction mixtures were incubated no. of rotamer outliers 0 0 at 30 °C for 30 min. Persulfide yields were quantified by using PDB entry 6B2M 6B2O a cold cyanolysis procedure,22 with yields typically of ∼30% a Data for the highest-resolution shell are shown in parentheses. relative to the starting disulfide concentrations. Persulfide b ∑ ∑ | − ⟨ ⟩| ∑ ∑ Rmerge = hkl j Ij(hkl) I(hkl) / hkl jIj(hkl), where I is the fi fl c ∑ − 1/2∑ | − preparations were used immediately without further puri ca- intensity of re ection. Rpim = hkl[1/(N 1)] j Ij(hkl) − ° ⟨ ⟩| ∑ ∑ tion or frozen at 20 C under argon for later use. I(hkl) / hkl jIj(hkl), where N is the redundancy of the data set. Dha μ d ffi e LarE (2.8 M, 0.087 mg/mL) was incubated in 20 mM CC1/2 is the correlation coe cient of the half-data sets. Rwork = ∑ || | − | || ∑ | | Tris-HCl buffer (pH 7.5) containing 150 mM NaCl, 20 μg/mL hkl Fobs Fcalc / hkl Fobs , where Fobs and Fcalc are the observed and μ D calculated structure factors, respectively. Rfree is the cross-validation R glucose oxidase, 2 g/mL catalase, and 0.3% (w/v) -glucose factor for the test set of reflections (5% of the total) omitted during (to ensure anoxic conditions), with various additives as model refinement. specified. Selected mixtures included 10% cell-free lysates from L. plantarum cultures that were induced with lactic acid grown without being shaken at 37 °C in MRS medium or non-induced. Some mixtures included 2 mM MgCl2 and 1 (Sigma) supplemented with 0.1% (v/v) Tween 20. At an mM ATP. Individual sulfur sources and a combined mixture of ∼ OD600 of 1, the cultures were induced by adding 200 mM sulfur compounds (1 mM cysteine, 1 mM glutathione, 1 mM sodium L-lactate as a dry powder or left uninduced and 3-mercaptopyruvate, 1 mM sodium thiosulfate, 1 mM sodium incubated for a further 2 h. The cells were harvested by sulfide, and 200 μM CoA) were tested. In addition, IscS (4.6 centrifugation at 5000g, and the pellets were washed in 100 μM, 0.21 mg/mL) with 1 mM cysteine was examined. For all mM Tris-HCl buffer (pH 7.5) containing 150 mM NaCl and samples, the reaction mixtures were layered with mineral oil stored at −80 °C. The cells (∼500 μL of cell paste) were and allowed to incubate at room temperature with 1 mL time mixed with glass beads (∼500 μL, mean diameter of 100 μm) points taken at 1 h or overnight. The samples lacking cell-free in 20 mM Tris-HCl (pH 7.5) containing 150 mM NaCl and, lysates were directly analyzed by mass spectrometry, but those for maintaining anaerobic conditions,19 20 μg/mL glucose amended with lysates were loaded by gravity flow at 4 °C onto oxidase, 2 μg/mL catalase, and 0.3% (w/v) D-glucose in 1.5 columns containing 200 μL of Strep-tactin-XT resin; the mL screw cap tubes (deaerated by blowing argon over the columns were washed using three 1 mL additions of 20 mM surface) and lysed by using a Mini-Bead Beater 16 (BioSpec) Tris-HCl buffer (pH 7.5) containing 150 mM NaCl, and the

5515 DOI: 10.1021/acs.biochem.8b00601 Biochemistry 2018, 57, 5513−5523 Biochemistry Article

LarE protein was eluted using six 250 μL additions of the same buffer containing 50 mM biotin neutralized with 50 mM NaOH. These protein samples were concentrated to at least 1 μM LarE using 10 kDa cutoff Amicon centrifuge columns prior to further analysis. Mass Spectrometry of LarE. The purified LarE protein samples (10 μLinbuffer), including those from the LarE regeneration studies, were injected onto a cyano-chemistry high-performance liquid chromatography column that was equilibrated in 0.1% formic acid and eluted with an increasing gradient of acetonitrile. The fractions were analyzed by electrospray ionization mass spectrometry (ESI-MS) using a XEVO G2-XS instrument in positive ionization mode. The protein masses were derived from the MS data using MaxEnt (Waters Corp.). LarE samples were also purified from Lc. lactis NZ3900 pGIR172 that had been induced for different lengths of time (1, 2, 3, and 4 h) and analyzed by the same ESI-MS approach. LarE Enzyme Assay. The activity of LarE was assessed indirectly by a four-step process using purified components and assaying for Lar activity. The first step was the production of P2CMN by a 3 h incubation of LarB (3 μM) with 1 mM ° NaAD in 100 mM NaHCO3 with quenching by heat (80 C for 5 min). The second step involved LarE synthesis of P2TCM; the step 1 mixture (4% of the final volume) was incubated with 10 μM LarE, 2 mM reducing agent [Cys, − − − fi β Figure 2. ESI-MS of (A) the CoA S S Cys176 disul de using dithiothreitol (DTT), CoA, -mercaptoethanol, glutathione, or D30A LarE and (B) the C176A variant of LarE, both purified in the ascorbic acid], 2 mM ATP, and 20 mM MgCl2 in 100 mM presence of CoA. Tris-HCl (pH 7) for 5 min at room temperature, and then the reaction was stopped by heat. Step 3 involved synthesis of the NPN cofactor and generation of active LarA by reacting the LarE was isolated from both Lc. lactis and E. coli using step 2 mixture (10% of the final volume) with 1.5 μM LarC, purification buffers containing CoA, and each protein was 1.5 μM T. thermosaccharolyticum LarA apoprotein, 1 mM crystallographically characterized. All LarE chains of the MnCl2, 0.1 mM CTP, and 40 mM D-lactate in 100 mM Tris- crystallographic hexamers had CoA bound. The phosphate HCl (pH 7) for 5 min at room temperature, with quenching by group of the 3′-phosphoadenosine portion of the molecule was heat. The final step was to measure L-lactate with an enzymatic found at the PP loop (the ATP ) of LarE (Figure kit (Megazyme). 3A). The adenine moiety forms two hydrogen bonds with the backbone atoms of Ala52 on strand β2. The ribose is stabilized β ■ RESULTS via interactions with the backbone atoms of Ala24 on 1 and Gly123 on β4. The diphosphate group of CoA was well- Structural Investigation of Cys176/Coenzyme A ordered in all structures but exhibited slightly different Interaction. Prior studies had demonstrated that the addition orientations between chain C and all other chains (Figure of CoA stimulates the activation of LarA apoprotein in 3B). The differences were more notable for the remaining mixtures containing P2CMN (the product of LarB), Mg·ATP, pantoic acid-alanine-cystamine portions of CoA; this part of 8 LarE, LarC, and LarA apoprotein. Unpublished observations the molecule is less ordered in all chains, does not show direct showed a 765 Da adduct present in LarE purified from Lc. contacts with the protein, and shows slight variations in chains lactis cells, which we presumed to be CoA. When LarE (as the A, B, and D−F (F is shown as a representative example) Strep II-tagged species lacking the N-terminal Met residue; compared to the very different orientation in chain C (Figure 31550 Da) was purified in the presence of CoA, this adduct 3B). The distinct orientations of the coenzyme paralleled the (32316 Da) became the majority species and CoA was shown differences in side chain orientations of the key residue, to form a disulfide with the single cysteine residue of LarE Trp97.9 Although MS experiments had previously indicated a (Cys176).8 Several thiol reductants (sulfide, Cys, DTT, and disulfide bond between CoA and Cys176 of LarE,8 no disulfide glutathione), including CoA, could reduce the disulfide and bond was detected in the electron density maps. This result confer LarE activity in the Lar assay, but only CoA provided suggested that photoreduction of the disulfide may have longer-term (20 h) stability to the protein (data not shown). occurred during data collection. Notably, the CoA sulfur atom An analogous CoA-protein disulfide (32274 Da) was formed was in the vicinity of the Cys176 side chain, with the two with the D30A variant of LarE (31508 Da) but not with the different orientations having distances of 8.8 and 10.4 Å C176A variant (31520 Da) (Figure 2). CoA is not required for between the Cys176 and the CoA sulfur atoms. The transition E. coli-derived LarE to convert P2CMN to P2TMN.9 The between the two orientations of the least ordered portion of CoA-protein disulfide form of LarE converts to LarEDha in the CoA would place the CoA sulfur close enough to the Cys176 LarA apoprotein activation mixture. While it is clear that CoA side chain to form the disulfide. interacts with LarE, further studies were required to establish The binding mode of the 3′-phosphoadenosine ribose the details of this interaction. portion of CoA is nearly identical to that of ATP, the

5516 DOI: 10.1021/acs.biochem.8b00601 Biochemistry 2018, 57, 5513−5523 Biochemistry Article

α 16 − Figure 3. Binding of CoA to LarE (PDB entry 6B2M). All comparisons are based on a C alignment of the full length protein. (A) The 2FO FC map of CoA bound to chain F is shown as blue mesh at 1 σ. Interacting backbone residues are illustrated as ribbons with secondary structure features labeled, and the PP-loop residue side chains are shown. Hydrogen bonds are indicated as red dashes. (B) Comparison of CoA binding orientations in LarE. Chain F is colored magenta (representative for chains A, B, D, and E) and compared to chain C (gray). The corresponding side chains of Trp97 and Cys176 are illustrated in the same colors with distances between the CoA and Cys176 sulfur atoms shown as black dashes. (C and D) Comparison of CoA and ATP (PDB entry 5UDS)9 binding to LarE. Interacting residues are shown in stick mode with hydrogen bonds indicated by red dashes. The carbon atoms of the CoA- and ATP-bound structures are colored magenta and cyan, respectively. The ligand phosphorus atoms of CoA are colored orange, and those of ATP are colored black. difference being that the triphosphate of ATP is positioned to (Figure 4). This result suggests that CoA binding is partly place the γ-phosphate into the PP loop, whereas it is the ribose inhibitory with respect to LarE activity. Dha phosphate group of CoA that is tightly associated with the PP- Additional Characterization of LarE . Strep-tagged loop residues (Figure 3C,D). LarE was expressed in Lc. lactis in the context of the entire lar fi 8 CoA Weakly Inhibits LarE Activity. ATP likely has a operon and puri ed. As previously described, nearly all of the binding affinity for LarE greater than that of CoA, compatible resulting LarE lacks its Cys176 sulfur atom according to ESI- MS (Figure 5A). This major form of the protein was observed with its greater number of interactions involving the PP loop, ff for samples isolated from cultures that had been induced from thus overcoming any potent inhibitory e ect of the coenzyme. 1 to 4 h (data not shown), so LarEDha is stable in the cells. Nevertheless, the apparent LarE activity was lower when using Along with LarEDha (31517 Da) the spectrum reveals likely β CoA compared to other reductants (Cys, DTT, -mercaptoe- sodium complexes of the protein (≥31540 Da); the intensity thanol, glutathione, or ascorbic acid) when the protein was of these forms was greatly reduced using an ammonium subjected to short-term treatment with these reagents and bicarbonate buffer. In contrast to the native LarE protein that analyzed using the indirect LarA apoprotein activation assay exhibits DTNB reactivity accounting for 0.93 Cys thiol per

5517 DOI: 10.1021/acs.biochem.8b00601 Biochemistry 2018, 57, 5513−5523 Biochemistry Article

Figure 4. Inhibitory effect of CoA on the apparent LarE activity compared to other reductants. P2CMN was produced enzymatically from NaAD, converted to P2TMN by LarE in the presence of the indicated reductants (2 mM) along with 2 mM ATP and 20 mM ff MgCl2 in 100 mM Tris-HCl bu er (pH 7), transformed into the NPN cofactor by LarC, incorporated into T. thermosaccharolyticum LarA apoprotein, and assayed for Lar activity, shown as the percent relative to the control without thiol. protein, no 412 nm peak was detected in the LarEDha protein. As further support for the presence of Dha, the addition of TCEP to LarEDha led to partial modification to form a species at 31767 Da, or 250 Da larger than LarEDha (Figure S1), consistent with the known reactivity of TCEP toward dehydroalanine,18 whereas this reagent had no effect on the mass spectrum of LarE. Regeneration of Cys-Containing LarE from LarEDha by CoA Persulfide. Regeneration of native LarE from LarEDha was achieved by using the persulfide of CoA followed by Figure 5. Regeneration of native LarE from LarEDha. (A) ESI-MS of reduction with DTT (Scheme 1). Incubation of LarEDha with LarEDha. (B) LarEDha treated with CoA persulfide (150 μM, 1 h at CoA persulfide (Figure 5B) resulted in substantial conversion room temperature and then overnight at 4 °C), revealing conversion of the Dha-containing protein (31517 Da) to a species (32316 to the LarE−S−S−CoA disulfide. (C) Sample from panel B reduced Da) consistent with the LarE−S−S−CoA disulfide. In with DTT (10 mM, 30 min at room temperature) to generate native Dha fi addition, we noted a minor species (32285 Da) suggesting LarE. (D) LarE treated with Cys persul de. (E) Sample from panel − − D treated with DTT. The Δ values are relative to the 31517 Da value the formation of the LarE S CoA thioether. Consistent with Dha the thioether interpretation, the same species formed for of LarE . The y-axes indicate the percent relative to the intensities of the maximum peaks. LarEDha treated with CoA rather than the CoA persulfide (Figure S2). Subsequent reduction of the CoA persulfide Scheme 1 mixture by DTT (Figure 5C) eliminated the putative disulfide species but not the proposed thioether and generated a peak (31551 Da) corresponding to native LarE. In contrast to the robust regeneration observed using the CoA persulfide, the Cys persulfide produced much lower levels of the LarE−S−S− Cys disulfide [31669 Da (Figure 5D)] and corresponding smaller amounts of the native LarE after reduction (Figure 5E). DTT reduction of the Cys persulfide-treated sample led to an increase in the 31669 Da species that we attribute to native LarE from LarEDha was demonstrated by these in vitro formation of the thioether between DTT and LarEDha, which experiments, it remains unclear whether a similar process happens to possess the same mass as the LarE−S−S−Cys occurs within L. plantarum cells. disulfide, as confirmed by a control reaction of DTT and the Alternative efforts to reincorporate sulfur into the LarEDha Dha-containing protein (Figure S3). No reactivity was protein were unsuccessful. These included using a variety of in detected between the glutathione persulfide and LarEDha vitro conditions, a range of potential sulfur donors, lysates from (data not shown). These results suggest the CoA binding native L. plantarum cells (induced by addition of L-lactate and fi fi site of LarE is used to speci cally bind the CoA persul de non-induced controls), other potential cofactors (MgCl2 and during the sulfuration reaction. Although regeneration of ATP, CoA, and glutathione), and cysteine and cysteine

5518 DOI: 10.1021/acs.biochem.8b00601 Biochemistry 2018, 57, 5513−5523 Biochemistry Article desulfurase, a known sulfur source for several other biological R212A) and the active D231R variant but had success only sulfur-containing components. In none of these cases did we with the C176A LarE sample. The W97A variant, thought to observe conversion of LarEDha to the WT species. be defective in its interaction with the P2CMN substrate,9 was Discovery of an NPN Adduct of LarEDha. Some poorly produced, thus preventing crystallization. The S180A preparations of LarEDha possessed a previously undescribed and R212A variants, considered to lack the P2CMN or species (31968 Da) that was 450 Da larger than the Dha- inorganic phosphate binding site, were examined using many containing protein (31518 Da) (Figure 6). Interestingly, this conditions, but the lack of successful crystallization suggests that a bound phosphate/sulfate molecule at this binding site is required for crystal formation. Although the D231R variant is as active as the WT enzyme, we observed no crystal formation, suggesting that the LarE trimer/hexamer interface where Asp231 is located may be disturbed, thus preventing crystal growth. The successful case of the C176A variant protein yielded the same crystal morphology and size as the WT enzyme. The overall fold of the C176A LarE structure (Figure 7A) was identical to that of the WT protein. Even the flexible linker between domains on which Cys176 is located matched in both structures. In addition, the bound phosphate previously noted in the WT enzyme9 was also present in the C176A structure,

Dha indicating that hydrogen bonding with the sulfur atom (Figure Figure 6. Example ESI-MS spectrum of LarE with an inset 7B) is not crucial for binding the phosphate of P2CMN. The depicting an expanded view of the region from 31900 to 32000 Da. loss of activity of the C176A variant must be based solely on the functional role of the thiol group as a sulfur donor as the species is consistent with the formation of a covalent adduct C176A structure shows no other structural difference from the between NPN and LarE. Such a species explains a previous WT enzyme. confounding observation related to LarA activation. Namely, We previously characterized several LarE variants of highly the incubation of purified LarA (1.4 pmol) with LarE (280 conserved residues to show the importance of the active site pmol, isolated from cells containing LarB and LarC) was cysteine (C176A), the PP loop (D30A), the P2CMN shown to result in activation of ∼30% of the lactate racemase phosphate binding site (S180A and R212A), and several key apoprotein.5 We interpret this prior result as arising from a residues in catalysis (W97A, R181K, and E200Q) while also secondary pathway for LarA activation that utilizes this new ruling out the involvement of other residues (E61Q and species with NPN covalently attached to LarE. D231R) (Figure 7B).8,9 Here, we characterized three addi- Structural Characterization of C176A LarE and tional variants to gain an even better understanding of LarE’s Functional Characterization of LarE Variants. We function. Glu223 is buried in a manner similar to that of attempted to crystallize several LarE variants with largely Glu200 in the C-terminal head domain of LarE, which forms reduced or abolished activity (W97A, C176A, S180A, and trimer/hexamer interactions and binds the phosphate moiety

Figure 7. Analysis of LarE variants. (A) Comparison of the six chains of WT LarE colored cyan (PDB entry 5UDQ9) and those of the C176A LarE variant colored dark red (PDB entry 6B2O). The Cys176 side chain and nearby phosphate molecule are shown. (B) P2CMN/AMP-bound model of LarE based on chain A of NMN-bound LarE (PDB entry 5UDR) and chain C of AMP-bound LarE (PDB entry 5UDT).9 Carbon atoms of the P2CMN model are colored gray, those of AMP magenta, and the side chains of residues that are altered in inactive variants dark red [D30A and C176A;8 W97A, C176A, S180A, R181K, E200Q, and R212A;9 and D128A (this publication)]. In addition, the side chains of residues that were substituted in active variants are colored black [E61Q and D231R9 and K101A and E223A (this publication)]. Hydrogen bonds are shown as red dashes, and important interactions are illustrated as black dashes with the distances indicated. (C) Activity results of three variants characterized in this publication.

5519 DOI: 10.1021/acs.biochem.8b00601 Biochemistry 2018, 57, 5513−5523 Biochemistry Article

Figure 8. Biosynthesis of the NPN cofactor for LarA apoprotein activation, regeneration of LarEDha, and an alternate pathway to generate lactate racemase activity. In the predominant pathway, LarB carboxylates nicotinic acid adenine dinucleotide (NaAD) and hydrolyzes its phosphoanhydride linkage, thus forming P2CMN. LarE activates the substrate carboxyl group by adenylylation, forms a thioester linkage with the substrate, and sacrifices its Cys176 sulfur atom, and then a second LarE repeats the cycle to produce P2TMN (blue arrow). LarC inserts nickel into this species to generate the NPN cofactor that activates LarA. LarEDha can be regenerated to native LarE by incubation with CoA persulfide and reductant (orange arrows), but it is unclear whether these reactions are physiologically relevant. A newly identified NPN-LarE intermediate functions in an alternative, minor pathway (yellow arrow) in which LarC installs nickel into the pincer cofactor while it is still bound to LarE. NPN is released from this purified protein for LarA activation. of the substrate, P2CMN. Furthermore, Glu223 forms the importance of the Glu200 side chain. Second, we examined hydrogen bonds to the nearby Arg214 and Arg221 residues, Lys101 that is conserved in LarE sequences and forms a salt with Arg214 being directly involved in phosphate binding. The bridge with AMP (Figure 7B). Surprisingly, the K101A variant E223A variant exhibited no significant change in activity was as active as the WT enzyme, suggesting that the lysine is compared to the WT enzyme. This finding shows that the not critical for stabilizing AMP binding and that other residues interaction of Arg214 with phosphate is unlikely to require on a nearby flexible loop might compensate for its absence Glu223 so that phosphate binding requires only Ser180 and through other AMP phosphate interaction residues.9 Asp128 is Arg212 (Figure 7B). The Arg214/Glu223 interaction is another highly conserved residue on the flexible loop, but this structurally similar to the nearby Arg181/Glu200 interaction, residue does not directly interact with AMP in the crystal but in that case, the E200Q variant was inactive, highlighting structure (Figure 7B). At a distance of 6.2 Å from the AMP

5520 DOI: 10.1021/acs.biochem.8b00601 Biochemistry 2018, 57, 5513−5523 Biochemistry Article phosphate, it is unclear what role this residue could play; nevertheless, it must be important as the D128A variant was inactive. This result demonstrates a key role for the flexible loop containing Asp128, either in its disordered state found in most structures or as the helical fold in the AMP-bound structure. ■ DISCUSSION NPN Biosynthesis, LarEDha Regeneration, and a New Pathway for LarA Activation. We combine our new results with those from prior studies in a summary depiction of the LarA apoprotein activation process (Figure 8). In the primary pathway, LarB converts NaAD to P2CMN,8 two molecules of LarE catalyze AMPylation of the pyridinium carboxyl groups followed by sacrificial sulfur transfers from Cys176 to form P2TMN (generating two copies of LarEDha),8,9 LarC inserts nickel by a CTP-dependent process,8,10 and the resulting free NPN activates LarA (Figure 8, blue arrow pathway). We have now shown that LarEDha can be recycled to regenerate native LarE by in vitro incubation with CoA persulfide and reductant (Scheme 1 and Figure 8, orange arrows). The CoA persulfide is likely to bind at the CoA binding site of LarE for this reaction because other tested persulfides were inefficient or not effective. Such binding Figure 9. Comparison of active sites of structurally related sulfur would perfectly position the persulfide to react with Dha. . (A) LarE from L. plantarum colored cyan (PDB entry Persulfides are known to have increased nucleophilic character 5UDR). (B) Homology model of LarETt from T. thermosacchar- compared to thiols and are of increasing interest in olyticum colored gray. (C) 2-Thiouridine synthetase (TtuA) from Th. biology.23,24 CoA persulfide is found in cells25,26 and has thermophilus colored light green (PDB entry 5B4E). (D) TtuA from been shown to bind tightly to short chain acyl-CoA P. horikoshii colored dark green (PDB entry 5MKP). (E) Adenosine 5′-phosphosulfate reductase from Ps. aeruginosa colored blue (PDB dehydrogenases.27,28 Although CoA persulfide is an attractive Dha entry 2GOY). (F) tRNA 2-thiouridylase (MnmA) from E. coli colored candidate for recycling LarE , the near exclusive presence of orange (PDB entry 2DEU). (G) tRNA 4-thiouridine synthetase this form of the protein in Lc. lactis cells containing pGIR172 (ThiI) from Thm. maritima colored purple (PDB entry 4KR6). (H) suggests this reaction is not physiologically relevant and LarE is Functionally related sacrificial sulfur insertase (THI4p) from S. a single-turnover enzyme. cerevisiae colored brown with bound adenylylated thiazole (PDB entry Our finding of what appears to be an NPN-LarE adduct 3FPZ). suggests that LarC can insert nickel into the incomplete pincer ligand that is still bound in the thioester linkage to LarE non-[4Fe-4S] cluster sulfide atom (positioned where a fourth (Figure 8, yellow arrow). The flexibility of the LarE loop cysteine is found for many similar clusters), which must be containing Cys176 could reasonably allow the pincer species to replaced for each round of synthesis.29 Structural similarities be metalated by LarC. The NPN-LarE species is presumed to and sulfur transfer involving a labile extra cluster sulfide were release NPN, accounting for the previously reported activation noted in a study of TtuA from Pyrococcus horikoshii that also of LarA apoprotein by just LarE,5 resulting in the generation of possesses a [4Fe-4S] cluster bound by three cysteines (Figure the LarEDha form of the protein. The in vivo relevance of the 9D).30 Furthermore, the structure of the more distantly related NPN-LarE adduct remains to be clarified; this species may be ortholog adenosine 5′-phosphosulfate reductase from Pseudo- abundant within the cell but converts to LarEDha as the protein monas aeruginosa also reveals a [4Fe-4S] cluster at this site;31 is purified. however, in that case, four protein cysteine residues are present LarE Homologues: Potential Roles of Additional and the cluster is not thought to donate a sulfur atom (Figure Cysteine Residues. We had previously noted that many 9E). Rather, this enzyme uses its [4Fe-4S] and flavin to LarE homologues have sequences in which Cys176 appears to reversibly transfer electrons from sulfite and AMP while be shifted by one residue and points toward two additional forming adenylyl sulfate and a reduced electron acceptor. cysteine residues, not present in the L. plantarum LarE Interestingly, two other structurally related sulfur transferases, sequence, one of which substitutes for Trp97 (Figure 9A,B).9 tRNA 2-thiouridylase or MnmA from E. coli (Figure 9F)32 and We speculate that these alternative LarE versions might tRNA 4-thiouridine synthetase or ThiI from Thermotoga catalyze a different mechanism of sulfur transfer that does not maritima (Figure 9G),33 also contain their catalytic cysteine require sacrifice of a side chain sulfur atom as observed in the residues in the same region as LarE’s Cys176. In these cases, benchmark LarE. Comparison of the location for these cysteine cysteine persulfides are generated, with disulfide exchange residues in a model of T. thermosaccharolyticum LarE (Figure involving another cysteine residue allowing for the insertion of 9B)9 to the structure of Thermus thermophilus TtuA,29 another a sulfur atom into the substrate. Although the structure has not PP-loop pyrophosphatase family member that catalyzes sulfur been obtained, ThiI from Methanococcus maripaludis was transfer for 2-thiouridine synthesis, reveals the similarity of the shown to possess a [3Fe-4S] cluster that was needed for sulfur LarE ortholog and the [4Fe-4S] cluster binding site bound by transfer.34 three cysteine residues in TtuA (Figure 9C). In that case, the One precedent exists for a sacrificial sulfur transferase, sulfur transferred to uridine is suggested to derive from the THI4p that functions in the thiamine biosynthesis pathway of

5521 DOI: 10.1021/acs.biochem.8b00601 Biochemistry 2018, 57, 5513−5523 Biochemistry Article

Saccharomyces cerevisiae.35 Although related in function to LarE this analysis, we speculate that some orthologs of LarE may and similarly thought to catalyze a single turnover, THI4p bind a [4Fe-4S] cluster providing them with the ability to act possesses a distinct fold (Figure 9H). Of further interest, catalytically. Our supposition of two forms of an enzyme, one Methanococcus jannaschii possesses an ortholog of THI4p that acting stoichiometrically to form LarEDha and the other acting uses exogenous sulfide for thiamin biosynthesis and is catalytically, has precedence in thiamin biosynthesis. catalytic.36 This example, like that mentioned above for ThiI, illustrates the situation in which orthologs of the same protein ■ ASSOCIATED CONTENT can exhibit distinct properties. In this case, one example is a * fi S Supporting Information sacri cial single-turnover enzyme while the other ortholog The Supporting Information is available free of charge on the exhibits multiple-turnover catalytic activity. ACS Publications website at DOI: 10.1021/acs.bio- LarE is structurally related to other sulfur transferases that chem.8b00601. contain [4Fe-4S] clusters or generate persulfides at their active sites, such as TtuA, MnmA, and ThiI. In addition, it is Mass spectrometric results for the interaction of TCEP, functionally related to sulfur transfer in thiamin CoA, and DTT with LarEDha (Figures S1−S3, synthesis that utilize two distinct mechanisms, one stoichio- respectively) (PDF) metric and one catalytic. Thus, we hypothesize that other LarE orthologs may bind a [4Fe-4S] cluster or generate persulfides and use these species for catalytic sulfur transfer. LarE ■ AUTHOR INFORMATION homologues that use three active site cysteine residues to Corresponding Author bind a [4Fe-4S] cluster could use the nonligated sulfur atom *E-mail: [email protected]. fi for attack of the AMPylated substrate without sacri cing a ORCID cysteine residue. Similarly, a persulfide could attack the Matthias Fellner: 0000-0003-3192-6984 adenylylated substrate with another cysteine forming a disulfide to allow for sulfur transfer. The active form of the Joel A. Rankin: 0000-0002-9775-4508 enzyme could then be reconstituted by the normal cellular Benoît Desguin: 0000-0002-5004-6918 cluster synthesis machinery or by persulfide formation to allow Jian Hu: 0000-0001-6657-9826 for additional rounds of sulfur transfer. It will be interesting to Robert P. Hausinger: 0000-0002-3643-2054 assess whether LarE functions catalytically in the multi-Cys Present Address ⊥ proteins. M.F.: Department of Biochemistry, University of Otago, P.O. Box 56, Dunedin 9054, New Zealand. ■ CONCLUSIONS Author Contributions We elucidated the structure of the complex between CoA and M.F. and J.A.R. contributed equally to this work. LarE and unexpectedly observed strong similarities in binding Funding modes for CoA and ATP. The sulfur atom of CoA can adopt This work was supported by the National Science Foundation multiple positions, including formation of the previously (Grants CHE-1516126 and CHE-1807073 to R.P.H. and identified disulfide with Cys176 of LarE, and the structures J.H.). provide insights into how CoA provides thermal stability to the protein. The persulfide of CoA was competent for restoring the Notes sulfur atom to LarEDha, whereas other LarE regeneration efforts The authors declare no competing financial interest. were poorly effective or unsuccessful. The low extent of LarE recycling in vitro and the lack of evidence for in vivo recycling ■ ACKNOWLEDGMENTS are consistent with our prior conclusion that LarE is a The authors thank Tony Schilmiller of the Michigan State 9 sacrificial protein. Only two LarE molecules are required for University Mass Spectrometry Core Facility for advice and activation of each cofactor that becomes bound to LarA, so a Eugene Mueller (University of Louisville, Louisville, KY) for stoichiometric mechanism may be sufficient for cellular needs. providing pBH402 encoding IscS. We identified a new species, the NPN adduct of LarE that explains the perplexing prior results in which purified LarE was ■ REFERENCES sufficient to activate LarA apoprotein. Release of NPN from (1) Juturu, V., and Wu, J. C. (2016) Microbial production of lactic the small proportion of LarE with a bound cofactor accounted acid: The latest development. Crit. Rev. Biotechnol. 36, 967−977. for LarA activation when using a vast excess of the accessory (2) Desguin, B., Soumillion, P., Hols, P., Hu, J., and Hausinger, R. P. protein. The observation of this adduct shows that LarC can (2017) Lactate racemase and its niacin-derived, covalently-tethered, catalyze the metalation reaction of the organic ligand bound in nickel cofactor. In Biological Chemistry of Nickel (Zamble, D., a thioester linkage to LarE. We determined the structure of the Rowinska-Zyrek, and Kozlowski, H., Eds.) pp 220−236, Royal C176A variant and found that the lack of the thiol side chain Society of Chemistry. ledtonoconformationalchangescomparedtothe (3) Fellner, M., Rankin, J. A., Hu, J., and Hausinger, R. P. (2017) conformation of the WT enzyme. We also created three Lactate racemase. In Encyclopedia of Inorganic and Bioinorganic additional active site variants of LarE and discussed their Chemistry, John Wiley & Sons, Ltd. activities in a structural context, along with those of previously (4) Desguin, B., Soumillion, P., Hausinger, R. P., and Hols, P. (2017) Unexpected complexity in the lactate racemization system of described variants. Finally, we performed a structural lactic acid bacteria. FEMS Microbiol. Rev. 41, S71−S83. comparison of the L. plantarum LarE to a homology model (5) Desguin, B., Goffin, P., Viaene, E., Kleerebezem, M., Martin- for an alternative version of LarE with a predicted cluster of Diaconescu, V., Maroney, M. J., Declercq, J. P., Soumillion, P., and cysteine residues and other related enzymes that possess [4Fe- Hols, P. (2014) Lactate racemase is a nickel-dependent enzyme 4S] clusters or a reactive persulfide in their active sites. From activated by a widespread maturation system. Nat. Commun. 5, 3615.

5522 DOI: 10.1021/acs.biochem.8b00601 Biochemistry 2018, 57, 5513−5523 Biochemistry Article

(6) Desguin, B., Zhang, T., Soumillion, P., Hols, P., Hu, J., and involving in hydrogen sulfide deteoxification. Biochemistry 54, 4542− Hausinger, R. P. (2015) A tethered niacin-derived pincer complex 4554. with a nickel-carbon bond in lactate racemase. Science 349,66−69. (26) Shen, J., Walsh, B. J. C., Flores-Mireles, A. L., Peng, H., Zhang, (7) Hausinger, R. P., Desguin, B., Fellner, M., Rankin, J. A., and Hu, Y., Zhang, Y., Trinidad, J. C., Hultgren, S. J., and Giedroc, D. P. J. (2018) Nickel pincer nucleotide cofactor. Curr. Opin. Chem. Biol. (2018) Hydrogen sulfide sensing through reactive sulfur species 47,18−23. (RSS) and nitroxyl (HNO) in Enterococcus faecalis. ACS Chem. Biol. (8) Desguin, B., Soumillion, P., Hols, P., and Hausinger, R. P. 13, 1610−1626. (2016) Nickel-pincer cofactor biosynthesis involves LarB-catalyzed (27) Shaw, L., and Engel, P. C. (1987) CoA-persulfide: a possible in pyridinium carboxylation and LarE-dependent sacrificial sulfur vivo inhibitor of mammalian short-chain acyl-CoA dehydrogenase. − insertion. Proc. Natl. Acad. Sci. U. S. A. 113, 5598−5603. Biochim. Biophys. Acta, Lipids Lipid Metab. 919, 171 174. (9) Fellner, M., Desguin, B., Hausinger, R. P., and Hu, J. (2017) (28) Tiffany, K. A., Roberts, D. L., Wang, M., Paschke, R., Mohsen, Structural insights into the catalytic mechanism of a sacrificial sulfur A.-W. A., Vockley, J., and Kim, J.-J. P. (1997) Structure of human insertase of the N-type ATP pyrophosphatase family, LarE. Proc. Natl. isovaleryl-CoA dehydrogenase at 2.6 Å resolution: Structural basis for − Acad. Sci. U. S. A. 114, 9074−9079. substrate specificity. Biochemistry 36, 8455 8464. (10) Desguin, B., Fellner, M., Riant, O., Hu, J., Hausinger, R. P., (29) Chen, M., Asai, S. I., Narai, S., Nambu, S., Omura, N., Hols, P., and Soumillion, P. (2018) Biosynthesis of the nickel-pincer Sakaguchi, Y., Suzuki, T., Ikeda-Saito, M., Watanabe, K., Yao, M., nucleotide cofactor of lactate racemase requires a CTP-dependent Shigi, N., and Tanaka, Y. (2017) Biochemical and structural cyclometallase. J. Biol. Chem. 293, 12303. characterization of oxygen-sensitive 2-thiouridine synthesis catalyzed (11) Krug, M., Weiss, M. S., Heinemann, U., and Mueller, U. (2012) by an iron-sulfur protein TtuA. Proc. Natl. Acad. Sci. U. S. A. 114, 4954−4959. XDSAPP: a graphical user interface for the convenient processing of (30) Arragain, S., Bimai, O., Legrand, P., Caillat, S., Ravanat, J. L., diffraction data using XDS. J. Appl. Crystallogr. 45, 568−572. Touati, N., Binet, L., Atta, M., Fontecave, M., and Golinelli- (12) Battye, T. G., Kontogiannis, L., Johnson, O., Powell, H. R., and Pimpaneau, B. (2017) Nonredox thiolation in tRNA occurring via Leslie, A. G. (2011) iMOSFLM: a new graphical interface for sulfur activation by a [4Fe-4S] cluster. Proc. Natl. Acad. Sci. U. S. A. diffraction-image processing with MOSFLM. Acta Crystallogr., Sect. D: − − 114, 7355 7360. Biol. Crystallogr. 67, 271 281. (31) Chartron, J., Carroll, K. S., Shiau, C., Gao, H., Leary, J. A., (13) Evans, P. R., and Murshudov, G. N. (2013) How good are my Bertozzi, C. R., and Stout, C. D. (2006) Substrate recognition, protein data and what is the resolution? Acta Crystallogr., Sect. D: Biol. − dynamics, and iron-sulfur cluster in Pseudomonas aeruginosa adenosine Crystallogr. 69, 1204 1214. 5′-phosphosulfate reductase. J. Mol. Biol. 364, 152−169. (14) Adams, P. D., Afonine, P. V., Bunkoczi, G., Chen, V. B., Davis, (32) Numata, T., Ikeuchi, Y., Fukai, S., Suzuki, T., and Nureki, O. I. W., Echols, N., Headd, J. J., Hung, L. W., Kapral, G. J., Grosse- (2006) Snapshots of tRNA sulphuration via an adenylated Kunstleve, R. W., McCoy, A. J., Moriarty, N. W., Oeffner, R., Read, R. intermediate. Nature 442, 419−424. J., Richardson, D. C., Richardson, J. S., Terwilliger, T. C., and Zwart, (33) Neumann, P., Lakomek, K., Naumann, P. T., Erwin, W. M., P. H. (2010) PHENIX: a comprehensive Python-based system for Lauhon, C. T., and Ficner, R. (2014) Crystal structure of a 4- macromolecular structure solution. Acta Crystallogr., Sect. D: Biol. thiouridine synthetase-RNA complex reveals specificity of tRNA U8 Crystallogr. 66, 213−221. modification. Nucleic Acids Res. 42, 6673−6685. (15) Emsley, P., Lohkamp, B., Scott, W. G., and Cowtan, K. (2010) (34)Liu,Y.,Vinyard,D.J.,Reesbeck,M.E.,Suzuki,T., Features and development of Coot. Acta Crystallogr., Sect. D: Biol. Manakongtreecheep, K., Holland, P. L., Brudvig, G. W., and Soll, Crystallogr. 66, 486−501. D. (2016) A [3Fe-4S] cluster is required for tRNA thiolation in (16) Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., archaea and eukaryotes. Proc. Natl. Acad. Sci. U. S. A. 113, 12703− Greenblatt, D. M., Meng, E. C., and Ferrin, T. E. (2004) UCSF 12708. Chimera–a visualization system for exploratory research and analysis. (35) Chatterjee, A., Abeydeera, N. D., Bale, S., Pai, P. J., Dorrestein, J. Comput. Chem. 25, 1605−1612. P. C., Russell, D. H., Ealick, S. E., and Begley, T. P. (2011) (17) Riddles, P. W., Blakeley, R. L., and Zerner, B. (1979) Ellman’s Saccharomyces cerevisiae THI4p is a suicide thiamine thiazole synthase. reagent: 5,5′-dithiobis(2-nitrobenzoic acid) – a reexamination. Anal. Nature 478, 542−546. Biochem. 94,75−81. (36) Eser, B. E., Zhang, X., Chanani, P. K., Begley, T. P., and Ealick, (18) Scian, M., Guttman, M., Bouldin, S. D., Outten, C. E., and S. E. (2016) From suicide enzyme to catalyst: The iron-dependent Atkins, W. M. (2016) The myeloablative drug busulfan converts sulfide transfer in Methanococcus jannaschii thiamin thiazole biosyn- cysteine to dehydroalanine and lanthionine in redoxins. Biochemistry thesis. J. Am. Chem. Soc. 138, 3639−3642. 55, 4720−4730. (19) Englander, S. W., Calhoun, D. B., and Englander, J. J. (1987) Biochemistry without oxygen. Anal. Biochem. 161, 300−306. (20) Mueller, E. G., Palenchar, P. M., and Buck, C. J. (2001) The role of the cysteine residues of ThiI in the generation of 4-thiouridine in tRNA. J. Biol. Chem. 276, 33588−33595. (21) Kabil, O., and Banerjee, R. (2012) Characterization of patient mutations in human persulfide dioxygenase (ETHE1) involved in H2S catabolism. J. Biol. Chem. 287, 44561−44567. (22) Wood, J. L. (1987) Sulfane sulfur. Methods Enzymol. 143,25− 29. (23) Cuevasanta, E., Möller, M. N., and Alvarez, B. (2017) Biological chemistry of hydrogen sulfide and persulfides. Arch. Biochem. Biophys. 617,9−25. (24) Park, C.-M., Weerasinghe, L., Day, J. J., Fukuto, J. M., and Xian, M. (2015) Persulfides: Current knowledge and challenges in chemistry and chemical biology. Mol. BioSyst. 11, 1775−1785. (25) Shen, J., Keithly, M. E., Armstrong, R. N., Higgins, K. A., Edmonds, K. A., and Giedroc, D. P. (2015) Staphylococcus aureus CstB is a novel multidomain persulfide dioxygenase-

5523 DOI: 10.1021/acs.biochem.8b00601 Biochemistry 2018, 57, 5513−5523